Back to search

University of Illinois at Urbana-Champaign

EMACS: efficient CXL-based memory access counters

Abstract

dc:description

This paper presents an efficient solution, EMACS, to count memory accesses for applications utilizing the CXL memory on an FPGA platform. CXL is a promising alternative to the mainstream DDR memory interface that can expand the capacity and bandwidth of a memory system. One type of CXL memory device comes in the form factor of an FPGA with a hard IP, providing extra programmability for its user and thus allowing the implementation of EMACS. EMACS consists of the following 3 key components. 1) A customized hardware logic on the FPGA that fulfills the counting mechanism. 2) A compatible kernel module to handle the communication to the logic. 3) A shell script that automates the control sequence to readily profile the memory accesses for any application. The mechanism and implementation of all three components will be elaborated in this paper, as well as the challenges met in the implementation process. In the end, the outcome of the EMACS tests on several benchmarks will be shown in the form of heatmap graphs.

Degree

thesis:*
Name thesis:degree_name
M.S.
Level thesis:degree_level
Thesis
Discipline thesis:degree_discipline
Electrical & Computer Engr
Grantor
University of Illinois at Urbana-Champaign
Year dc:date
2024

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Yu, Zeduo (Douglas)
Contributors dc:contributor
  • Kim, Nam Sung

Subjects

dc:subject × 3

Rights

dc:rights
Statement dc:rights
  • Copyright 2024 Zeduo (Douglas) Yu
Language dc:language
en, eng

Identifiers

dc:identifier.*
Handle dc:identifier
https://hdl.handle.net/2142/124609

Chain of custody

source
Harvested from
University of Illinois - Urbana-Champaign
Base URL
www.ideals.illinois.edu/oai-pmh
Last updated
2026-07-22
Source record
OAI-PMH GetRecord
citation

Yu, Zeduo (Douglas). EMACS: efficient CXL-based memory access counters. Thesis thesis, University of Illinois at Urbana-Champaign, 2024. https://hdl.handle.net/2142/124609